Abstract

In the novel part of this talk, I will present our proposed statistical methodology for studying the causal relationship between people’s opinions or knowledge about an issue (X) and their corresponding decisions or actions (Y). A simple example: X may represent an individual’s opinion on the efficacy of a particular vaccine, while Y is the decision to get vaccinated. Inferring causality in this context is challenging, especially because of the presence of unknown or unobserved confounders.

To address this, instrumental variables (IVs), a common tool in econometrics, are often employed. However, when examining the link between opinions and decisions, typical IV candidates may fail by acting as confounders themselves. We propose a novel IV specifically designed for these scenarios.

The first part of the talk will provide a concise overview of causal inference, covering all relevant concepts and terminology.

Joint work with Raanan Sulitzeanu-Kenan and Micha Mandel.